Blog

Organizations across multiple business sectors use data analytics in pursuit of real-time, in-depth understanding of major strategic and mission-critical factors, such as customer engagement, mobile platform usage and…well, restroom availability.

That’s certainly been the case at Starbucks over the past few months as the firm’s well-established digital DNA has been put to good use as the coffee giant found itself even more dependent on its mobile app to connect with customers as COVID lockdowns shuttered its stores. 

The company has relied on its data analytics capabilities to inform strategic and tactical responses to the virus and enable it to pivot operationally according to local circumstances – which is where the restrooms come in. Bathroom use was a metric that no-one really expected to be focusing on, explains Rajesh Naidu, the firm’s VP of Architecture, Data and Analytics Technology, but realisation dawned that it is actually very relevant,

Starbucks is not only a go-to destination for great beverages, food and experiences, but customers also know us for having clean bathrooms. So we had to develop a new way for store partners to quickly enter data about their respective stores, what amenities are open or closed and then surface this information in relevant reports and eventually bring this to our customers…We had to figure out a way of gathering bathroom data without putting in any kind of creepy things, like sensors and things like that.

A quick adaptation of the Tableau-powered daily top line dashboard report was completed in a matter of weeks, with Starbucks tech teams doubling down to bring data to their business colleagues. It was reflective of a whole new way of operating overall, says Naidu: 

We had to run our stores in different formats and modes, something that we had never done before. This whole notion of having the point of sale in the store, but turned off and actually having customer orders [go] through drive-thru or the curbside, was a new paradigm for us. We really needed to make this information discoverable for the customers, so that they know what mode a store is operating in when they come to Starbucks. 

A Tableau democracy

The COVID crisis has inevitably brought about changes in customer behavior that need to be understood and reflected in operational terms. For example, with workplaces closed, the early morning commuter coffee pick-up has largely gone away, while ‘Starbucks runs’ by people working remotely tend to come later in the day and are usually larger in size as orders are placed for everyone isolating at home. 

Naidu references five clear shifting customer behaviors that the firm’s data analytics work with Tableau has identified: 

• A fundamental need to be seen and experience a feeling of connection to others. 
• A seeking-out of experiences that effortlessly fit the lifestyle of customers. 
• An appreciation for consistent experiences is something customers really strive for. 
• A desire for high quality and sustainable products and experiences that support the wellbeing of the people and the planet. 
• Increasing loyalty to brands with strong values. 

Using data derived from Tableau, Starbucks has been able to react to changing circumstances with the necessary agility, as well as enabling its ‘partners’ – AKA employees – to deliver what Naidu calls “the human connection” in dealing with caffeine-thirsty customers. 

As a case in point, he cites the creation of that top line report on the state of the business mentioned earlier, developed from Tableau data by the Analytics Insight Team working in tandem with the Starbucks tech team, with the results shared across the company: 

Our main driver with this report is around how to get the right data points to the right leaders at the right time and to the right team. We refer to this as ‘data democratisation’ in our vocabulary here at Starbucks. The purpose of this report is to assist with what’s going on in our business and to provide data to the business partners so they can see the data holistically. 

Transparency of data sharing is one of Starbuck’s operating pillars, he says:

We make sure that we have data available to all levels of retail leadership as well as retail partners. This is really to give our retail partners a great sense of the business. Each one of them has accountability for the P&L for that store and we want to make sure that they have all the information they need at their hands. The daily top line report has specifics around site selection and other criteria that we also share internally, to make sure that our partners understand how the data comes together so they can make the right informed decisions.

The holistic approach Naidu cites was made possible in large part by Starbucks Azure-first strategy, which helped when migrating multiple data services to a single location. This data pooling was an iterative process across the company, he adds: 

We had user feedback, not just from our retail partners, but also from our non-retail partners in our support center in Seattle. The second thing was listening to our customers and actively understanding what our customers habits were going to be during the pandemic and post-pandemic. We got guidance from our systems teams, our analyst teams and even from the business teams. In addition to bringing the data that we had together, we had to also bring in data from external services. Making sure we added high quality, high fidelity data from external sources was something that we certainly needed. 

The foundation that we put together with an Azure-First approach to building our data lakes and data platform certainly helped us. Data governance, metadata management and taxonomy, all the things which we find sometimes to be boring, they are critical things in this exercise here, because we had to have a common definition across all of the teams to make sense of the data. And the last thing we focused on was data literacy. Even during these times, we put together training and learning journeys for our partners to help them understand the data and also work with these tools.

Future brewing

Looking ahead, Naidu reckons that Starbucks leaning into data analytics and understanding will help shape future tech-enabled innovation: 

We have some future capabilities, like hyper-personalization and radical automation. These are things which are not just buzzwords for us, but things which we are focusing on really to have data to make informed decisions and allow more time for our partners to have that human-to-human connection. 

And inevitably there’s an Artificial Intelligence (AI) and Machine Learning (ML) angle in play here, although Naidu has an important caveat here: 

I think before we get into AI and ML too far, it’s about getting the right data in the organization, being clear about what it is and who can use it and then taking it forward from there.

In terms of potential applications for AI tech, the the focus is on opportunities that really allow for greater human interaction, he says:

We want to see what can our customers expect for a more personalised digital experience, made possible by the Starbucks AI initiative we call Deep Blue, which is really tailored to look at where customers live, what they order, what their habits are and what the tendencies would be. 

So certainly for us, leveraging AI to deepen the digital relationship and architecting experiences that provide that experiential retail touch point to customers, is something where we want to surprise and delight our customers in new and different ways. We certainly are looking at use cases that align with our values . We want this to be cool, not creepy, technology for us. A lot of the AI/ML for us is based on a foundation of data democratisation.

That concept of democratisation remains hugely important, he adds: 

It’s important for our digital transformation. We want the technology teams to get out of the way and for the people who need to use the data to understand the data so they can analyse it. How can we enable some of those self-service capabilities and other things in our AI? It has to be explainable AI at the end of the day, so we are looking at ways of improving the partner experience. It’s not just customer experience that we’re focused on, but for our partners in our store. Are there tasks around inventory management or forecasting or labor scheduling, where we can take out some of the manual work and the guesswork and enhance the partner experience? We all need to be data-literate at Starbucks and be able to explain how we’re using AI.

Ultimately having a strong data-centric mindset will support further acceleration of digital transformation across Starbucks franchises and channel business, concludes Naidu: 

Today Starbucks has nearly 33,000 company-operated and licensed stores around the world; by 2030, we expect to reach approximately 55,000 stores and continue to develop experiences that address evolving customer routines. In the US, this will be a mix of new store formats and experiences, including drive-thru Starbucks and curbside pick-up, designed to meet customers where they want to engage with Starbucks. All this is possible due to the data foundation we have built and the ability to quickly surface development analytics and visualisations through the Tableau platform that we have here at Starbucks.